Apparatus and method for classifying regions of an image, based on the relative importance of the various areas and to adaptively use the importance information to allocate processing resources, e.g., bit allocation in an encoding environment.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A method for classifying a block within a current image of an uncompressed input image sequence, said method comprising the steps of: (a) mapping a chrominance component of a block in the current image of the uncompressed input image sequence in a chrominance plane; (b) defining a decision region by four straight lines in said chrominance plane; and (c) classifying said block as a skin-tone block if said chrominance component falls within said decision region.
2. The method of claim 1, wherein said defining step (b) defines said decision region by four straight lines expressed as: EQU C.sub.b.ltoreq.-5.7C.sub.r +227 EQU C.sub.b.gtoreq.-C.sub.r -10 EQU C.sub.r.gtoreq.3 EQU C.sub.b <-1 where C.sub.b is representative of an ordinate of said chrominance plane and C.sub.r is representative of an abscissa of said chrominance plane.
3. The method of claim 2, further comprising the step of: (c) applying said skin-tone block to generate an importance map.
4. The method of claim 3, wherein said applying step (c) further comprises the step of: (c1) using an edge block to generate said importance map.
5. The method of claim 3, wherein said applying step (c) further comprises the step of: (c1) using a motion block to generate said importance map.
6. The method of claim 3, wherein said applying step (c) further comprises the step of: (c1) using an importance map of a prior image to generate said importance map of the current image.
7. The method of claim 1, further comprising the step of: (d) using luminance information to qualify said classifying step (c).
8. A method for allocating coding bits to a block within an image of an uncompressed input image sequence, said method comprising the steps of: (a) obtaining an importance information for the block within the image, by receiving importance information from a block classifier that maps a chrominance component of said block in the image of the uncompressed input image sequence in a chrominance plane to detect a skin-tone block; and (b) allocating coding bits to said block in accordance with said importance information.
9. The method of claim 8, wherein said obtaining importance information step (a) further comprises the step of: receiving importance information from a block classifier for detecting an edge block.
10. The method of claim 8, wherein said obtaining importance information step (a) further comprises the step of: receiving importance information from a block classifier for detecting a motion block.
11. The method of claim 8, wherein said obtaining importance information step (a) further comprises the step of: receiving importance information from a block classifier using an importance map of a prior image.
12. The method of claim 8, wherein said allocating step (b) comprises the step of: (b1) allocating the coding bits in accordance with an importance factor that accounts for a significance of the block.
13. The method of claim 12, wherein said allocating step (1) allocates the coding bits in accordance with: EQU Q.sub.i =Q.sub.p (1+global_adj+local_adj).times.(Imp_factor) where Q.sub.i is a quantization scale for a current macroblock i, Q.sub.p is an average quantization scale over an entire previous frame, global_adj is a global adjustment, local_adj is a local adjustment, and Imp_factor is said importance factor for said current macroblock.
14. The method of claim 12, wherein said importance factor is expressed as: ##EQU3## where .epsilon. is a constant, .mu. is a constant, N.sub.rem is a number of remaining blocks, B.sub.i is a significance value of the ith block, and B.sub.curr is a significance value of the current block.
15. The method of claim 12, wherein said allocating step (b) further comprises the step of: (b2) allocating the coding bits in accordance with a global adjustment that accounts for bits used to encode a previous frame.
16. The method of claim 15, wherein said global adjustment is expressed as: EQU global_adj=(Bprev-T)/(2*T) where Bprev is a number of bits used to encode said previous frame, and T is a target bit rate for a current frame.
17. The method of claim 15, wherein said allocating step (b) further comprises the step of: (b3) allocating the coding bits in accordance with a local adjustment that accounts for a number of bits spent up to a present block.
18. The method of claim 17, wherein said local adjustment is expressed as: EQU local_adj=k*(actual_bits_spent_so_far-projection)/bit-rate, where k is a constant, actual_bits_spent_so_far is said number of bits spent up to the present macroblock, and projection is a number of bits that is projected to have been spent up to the present macroblock and bit-rate is a rate used to encode the image.
19. The method of claim 18, wherein said projection is expressed as: ##EQU4## where w.sub.i is a weight for a macroblock i and d.sub.i is a distortion for said macroblock i.
20. The method of claim 8, wherein said obtaining importance information step (a) comprises the step of: filling or purging said received importance information from said block classifier for detecting a skin-tone block.
21. An apparatus for allocating coding bits to a block within an image of an uncompressed input image sequence said apparatus comprising: a block classifier for generating an importance map for indicating an importance information of at least one block within the image, where said importance information is generated by mapping a chrominance component of said at least one block in the image of the uncompressed input image sequence in a chrominance plane to detect a skin-tone block; and a controller, coupled to said block classifier, for allocating coding bits to said block in accordance with said importance information.
22. The apparatus of claim 21, wherein said block classifier comprises: at least one detector for classifying a block as a skin-tone block, an edge block, or a motion block; and an importance map generator, coupled to said at least one detector, for generating said importance map.
23. A computer-readable medium having stored thereon a plurality of instructions, the plurality of instructions including instructions which, when executed by a processor, cause the processor to perform the steps comprising of: (a) mapping a chrominance component of a block in the current image of an uncompressed input image sequence in a chrominance plane; (b) defining a decision region by four straight lines in said chrominance plane; and (c) classifying said block as a skin-tone block if said chrominance component falls within said decision region.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
April 21, 1999
August 14, 2001
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.